Defect Image Characterization And Recognition Of Steel Butt Welds Based On Ultrasonic Phased Array

Welding connection structure have the advantages of high mechanical properties,good sealing performance, ease disassembly, light weight and short production cycle,has been widely used in the industrial field. Steel welding might have various internaldefects in the forming process, these defects can cause the performance of the steeldrops significantly. Ultrasound phased array can adjust the sound beam angle and depthof focus, in order to detect complex structures and blind spot detection of the location ofdefects. Due to its high sensitivity, speed, intuitive result, real-time display anddetection of complex-shaped surface or inaccessible parts, has been more and more usedin engineering detection. However Phased array also exist some problems in defectrecognition, the formation of some defects in the phased array image features is notobvious, it is difficult to study directly observation of its characteristics with the eye, itis necessary to identify by image feature extraction.According to "NB/T4701416-2011welding procedure qualification for pressureequipment and Welding specification for Pressure Vessel","JB/T4730-2005nondestructive testing of pressure equipment" standard, design defects specimen withpores, cracks, no penetration, lack of fusion and slag inclusions and so on. Specimendefects are qualitatived by ray detection method, in the case of known qualitativedefects, compare with defect detection results of ultrasonic phased array, studyultrasonic phased array inspection showing typical defects of images and featuresanalysis.In The article, combination A-scan and S-scan, ultrasound phased array detectionmethod can effectively identify the defects in butt welds, typical defect image obtainedby ultrasonic phased array detection and image feature analysis and characteristicsextraction. Common features can be divided into a gray feature, geometric shapefeatures and texture features. Using gray histogram of images in a variety of differentgray scale pixel distribution in probability and statistics and obtaining image grayscalecharacteristics; using Gaussian-Laplace defect image edge extraction and obtainingimage geometry characteristics; texture feature analysis and typical defects imageextraction base on GLCM, obtaining the image texture features. Finally, using statisticalmethods statistical characteristics of the typical defect images, summed up the unique characteristics of different typical defects. Studies have shown that the ray testingresults in accordance with the texture features. This paper presents the phased arrayimage’s texture feature extraction methods, combined with ray testing results. It caneffectively identify the defects in butt welds, detection sensitivity and detectionaccuracy compare with A-scan have greatly improved and Provide strong support forthe popularization and application of the phased array method.